yade-dev team mailing list archive

[Václav Šmilauer <eudoxos@xxxxxxxx>]

> The previous problem with "disappearing" particles was, that normalization
> of NULL-vector in WM3 gives (0,0,0), but Eigen gives (nan,nan,nan).
> So I needed to add such check in yademath. I do not know, what
> implementation is more correct: WM3 or Eigen.

Definitely Eigen's (since normalized vector should have |v|=1, right?). do NOT add this to yadeMath, it is correct to fail for such nonsense as normalizing 0 vector. Please document where it happens, that part of yade has to be fixed.